the athletes are not writers.” Only when subjects were allowed to draw dia-
grams to represent the relations expressed in the given statements could they ar-
rive at the correct logical conclusions.
The question of transfer is central to the assumption. What the research sug-
gests is that logical reasoning is situation specific, in which case it is not readily
transferable. But the ease with which we process simple syllogisms makes it
appear as though exercises in syllogistic reasoning will increase our logical
abilities overall. Furthermore, the history of grammar instruction, as well as the
folk psychology that informs much of what we do in education, inclines us to
believe not only that grammar is an exercise in logic but also that logical rea-
soning is as innate as breathing. If we can do it at all, we can do it an ywhere.
This is probably an illusion. As Johnson-Laird (1983) reported, no amount
of practice with syllogisms of the “all of the students are athletes” type makes
formulating a valid logical conclusion easier. It’s the equivalent of trying to
prepare for a marathon by running 50-yard dashes. Running is involved in both
cases, but 50-yard dashes will do little to prepare one for a marathon. On this
account, even if we accept the premise that grammar instruction exercises logi-
cal reasoning, we can predict that no amount of grammar study will have a sig-
nificant influence on students’ logical thinking in general. It will affect only
their logical thinking with regard to grammar. The situation-specific character-
istic of logical reasoning suggests that students may fully master grammar and
still reason illogically on a regular basis.^3
Furthermore, a wide range of research suggests that general logical reason-
ing is related to intelligence, which increasingly has been viewed not only as
the ability to develop multiple mental models to process experiences and solve
problems but also as the ability to select the best one consistently from among
the competing alternatives (Alcock, 2001; DeLoache, Miller, & Pierroutsakos,
1998; Herrnstein & Murray, 1994; Pinker, 2002; Rumelhart & McClelland,
1986; Steinberg, 1993).^4 Formal instruction, of course, does not have a signifi-
cant effect on intelligence (Pinker, 2002).
At this point, our analysis of the first two assumptions indicates that a signif-
icant disconnect exists between grammar instruction and learning outcomes.
The final assumption, that grammar instruction improves writing and reduces
22 CHAPTER 2
(^3) Following a suggestion by Bloom (1994), Pinker (2002) stated that “The logic of grammar can be
used to grasp large numbers: the expressionfour thousand three hundred and fifty-sevenhas the gram-
matical structure of an English noun phrase likehat, coat, and mittens. When a student parses the number
phrase she can call to mind the mental operation of aggregation, which is related to the mathematical op-
eration of addition” (p. 223). To the best of my knowledge there is no supporting evidence for this claim.
Also, what Pinker described here is merely a mnemonic, not a logical operation. 4
Although educators have thoroughly accepted Gardner’s (1983, 1993, 2000) theory of multiple
intelligences, the majority of scholars in psychology and cognitive science seem to have dismissed it,
largely on the grounds that the theory lacks empirical support (Klein, 1998; Morgan, 1996).